And yet, despite this abundance of analytics, decisions aren’t getting better. If anything, they’re getting slower.
Executives keep asking the same quiet question: Why isn’t all this expensive data helping us move faster?
The truth is simple. Most organizations have built the insight layer but never built the decision layer, the part that turns insight into consistent, high-quality actions.
Warren Powell, a Princeton professor who spent a career studying sequential decision-making, helps illuminate the gap. I hadn’t seen Powell since my grad student days, but now as an emeritus he is writing, online and off, perhaps more actively than ever before. One of his key messages is that analytics only creates value when it is woven directly into how decisions are made, not when it sits nearby like a well-meaning but passive bystander.
Until that decision layer exists, companies will continue producing more charts without producing more clarity.
The Expectation vs. The Reality
Executives tend to imagine a beautiful, linear flow: analytics produce an insight, the insight informs a decision, the decision produces a result.
A tidy little triangle. Symmetrical, elegant, and suitable for framing. The reality is less of a flowchart and more of a game of telephone.
Analytics produce a report, which sparks a debate, which triggers a delay, which eventually ends in an intuition-driven choice that looks suspiciously like what you would have done if you’d never hired a data team in the first place.
This isn’t because executives are ignoring the data. It’s because insights don’t automatically come with instructions. Without a decision layer, analytics are just… well… diagnostics. Informative, yes. Transformational, no.
To fix this, leaders need a simple way to recognize the kind of decision they’re making and what type of decision logic belongs there.
Powell gives us exactly that.
The Four Ways Your Organization Actually Makes Decisions
Powell’s academic framework arrives with intimidating names. But beneath the math are four familiar patterns, four ways organizations decide what to do next. Think of them as the Four Modes of Choosing.
1. The Rule of Thumb
These decisions run on simple, stable rules. When inventory hits ten units, reorder. When a customer’s engagement score drops below forty, flag them for review.
The Rule of Thumb works because it’s reliable, fast, and (crucially) repeatable. Every company has hundreds of these decisions hiding in plain sight.
2. The Here-and-Now
This mode optimizes the immediate moment without pretending to know the future. Dynamic pricing. Shifting media spend based on today’s performance. Adjusting call-center staffing for the afternoon rush.
It’s the business equivalent of tightening the sails based on the wind hitting your face right now.
3. The Long Game
Here we look beyond the quarter. This is where retention interventions, loyalty investments, and R&D bets live. Powell calls this a Value Function Approximation, which is a fancy way of saying: don’t eat the marshmallow yet.
A dollar spent on a customer today shows up as a cost now but as recurring revenue three quarters from now. The Long Game makes that trade-off explicit.
4. The Crystal Ball
This is the planning mode. Not prediction… planning. Supply chain scenarios. Budget cycles. Multi-quarter roadmaps.
The Crystal Ball isn’t about expecting one future; it’s about rehearsing several and choosing the strategy that survives them all.
Once leaders can spot which mode they’re in, decision clarity improves immediately. The team stops arguing about the wrong kind of logic and starts using the right one.
Where Most Analytics Fails
Most companies never reach this point. They build dashboards that describe the past. They build predictive models that describe the future. But nothing in their system tells them what to do.
That missing step—the decision—is where value leaks out. Without a decision layer:
- Teams make inconsistent choices.
- Meetings multiply.
- Response time slows to a bureaucratic crawl.
- Institutional knowledge lives in heads, not systems.
- Complexity scales faster than the organization can cope with it.
Analytics becomes something the company looks at, not something the company runs on.
What It Looks Like to Build a Decision Layer
You don’t need exotic AI or a dozen PhDs. You need structure.
It starts by identifying which type of decision each problem deserves. Churn management isn’t The Here-and-Now; it’s The Long Game. Inventory replenishment is a Rule of Thumb. Annual planning? Definitely the Crystal Ball.
Once the type is clear, analytics stop being decorative. They start generating actual candidate decisions.
- A low-inventory alert doesn’t just inform you; it generates a draft purchase order.
- A retention model doesn’t just score customers; it proposes an intervention and estimates the downstream value.
- A planning model doesn’t spit out a forecast; it plays out three futures and shows which strategy holds up across all of them.
And then something subtle but important happens: you begin measuring decisions instead of dashboards.
Instead of asking, “Did we report accurately?” you begin asking, “Did we choose wisely, given what we knew at the time?”
That simple shift transforms analytics from a reporting function into a learning system.
Seeing It in Practice
Consider churn analytics. Most companies stop at predicting who will leave. But the value lies not in spotting churn… it’s in changing it. That requires The Long Game: deciding whom to intervene with, how aggressively, and why.
Or look at inventory. In uncertain environments, the humble reorder point often beats sophisticated forecasting. The Rule of Thumb, when chosen deliberately, is often the smartest tool in the shed.
And strategic planning becomes far more resilient once you treat it as a rolling lookahead instead of a once-a-year ritual that feels more like tax preparation than strategy.
What Leaders Should Do Next
Here is your Monday morning mandate.
The next time someone brings you a dashboard, slide it gently back across the table and say, “This tells me what happened. Now show me the policy for what we do about it.”
Stop grading your analytics team on insight production. Start grading them on decision quality. And build a governance model where decisions, much like products, are continuously improved.
Data is not the bottleneck. Insight is not the bottleneck. Decision integration is the bottleneck.
Powell’s framework helps leaders fill that missing middle, turning analytics into something more valuable than information.
It becomes action. And that is where transformation begins.
Editor’s Note: See Powell’s article here: https://castle.princeton.edu/makingdecisions/.
For more columns from Michael Bagalman’s Data Science for Decision Makers series, click here (from All Things Innovation) and here (from All Things Insights).
